基于多源/多时相异质影像集成的滨海湿地演化遥感监测技术与应用研究

With the strengthening efforts from developing costal coastal zone, coastal wetlands have been severely disturbed by human activities, either in its quantity or quality. The effects from human activities in wetland evolution is an important content of "Impact of human activities on coastal ecosystems and environment" in the LOICZ study. This project aims for monitoring temporal evolution of wetlands and serving for wetland conservation, and carries on our study with three types of coastal zones including fully open type (Yancheng coastal wetland), open type (the south coastal wetland of Hangzhou Bay), semi-open type (Xiangshan coastal wetland). The project first studies fusion techniques of datasets from different sensors, with different spatial and temporal resolutions, heterogeneous remote sensing images covering different spectral interval, explores techniques of dimensionality reduction for high-dimensional fused feature dataset, and proposes extraction and recognition techniques of critical geographical features of wetlands. After that, this project makes a quantitative analysis in the relationship between the changes of coastal wetland shoreline and wetland development intensity, investigates the capacity of wetland resource development and utilization, studies the relationship between human disturbance activities with the evolution of different types of wetlands and reveals the succession rule of coastal wetland patterns and ecological functions along with its underlying driving mechanism under the influence of human services. The research result will greatly serve for the delineation of wetland ecological red line and wetland ecology and conservation.

随着海岸带开发力度的加强，滨海湿地无论是数量或质量都受到人类活动的严重干扰。人类活动影响下的湿地演化是LOICZ研究中“人类活动对近海生态系统与环境的影响”的重要研究内容。本项目以监测湿地时空演化、服务湿地生态保育为应用目标，以完全开敞型（苏北盐城滨海）、开敞型（杭州湾南岸）、半开敞型（象山港沿岸）滨海湿地为例，研发不同传感器、不同时空分辨率、不同波谱覆盖区间的异质遥感影像数据的融合技术、基于高维融合信息的数据快速降维技术、基于降维信息的湿地关键地理特征的提取和识别技术。在此基础上，定量分析滨海湿地岸线变迁及湿地开发强度之间的关系，探讨滨海湿地资源开发利用的承载力，研究滨海湿地人类干扰活动与不同类型湿地演化之间的相互关系及内在联系，揭示人类影响下的滨海湿地格局及生态功能演替规律及其深层驱动机制，服务于湿地生态红线划定与湿地生态保育。

本项目针对由遥感数据源信息单一、湿地解译智能化程度低引起的滨海湿地类型演替分析困难、时空变化驱动力难以定量刻画等关键问题，创新研发了不同传感器、不同分辨率遥感数据融合的湿地地理要素提取和识别技术，深入把握了滨海湿地岸线格局演化、湿地类型时空演替以及生态价值变化规律，进而实现了监测滨海湿地时空演化、服务湿地生态保育的应用目标。项目研究获得了如下成果：（1）研发了多源/多时相异质遥感影像的流行学习降维特征信息提取方法，为海岸带关键地理特征识别奠定数据基础；（2）构建了低维流型数据驱动下的海陆边界划分和湿地范围界定模型；（3）提出了基于遥感时间序列物候特征的滨海湿地地物精细分类策略，极大提高了滨海湿地类型识别精度；（4）分析了滨海湿地岸线变迁、格局演化及开发强度分析；（5）在滨海湿地分类的基础上，进行典型区域滨海湿地空间演替规律、生态风险演化与模拟研究；（6）最后将本项目方法与成果推广应用于人为干扰下的滨海湿地生态服务价值评估和湿地保育等工作。本项目研究成果在国内外学术刊物上发表论文71篇（其中SCI/SSCI 检索论文33篇），国内重要出版社出版专著5部，申请/授权国家发明专利13项，荣获浙江省和宁波市哲学社会科学优秀成果一、二奖，并多次获省部级领导肯定性指示。本项目培养博士毕业生1名，硕士研究生毕业生12名（多名研究生获浙江省优秀毕业生荣誉称号），支撑2位项目组成员入选浙江省特殊支持人才。研究显示，从多源/多时相的异质遥感影像集成的视角进行滨海湿地关键地理要素地提取和识别，有助于更加深入地探讨滨海湿地资源开发利用的承载力，揭示人类活动与不同湿地演化类型之间的相互关系及内在联系，厘清人类影响下滨海湿地格局及生态功能演替规律及其深层驱动机制，从而支撑滨海湿地生态红线划定与湿地生态保育工作。